This is the mail archive of the
mailing list for the GCC project.
- From: Eric Christopher <echristo at redhat dot com>
- To: Ulrich Weigand <weigand at i1 dot informatik dot uni-erlangen dot de>
- Cc: gcc-patches at gcc dot gnu dot org, uweigand at de dot ibm dot com
- Date: Mon, 03 May 2004 17:52:05 -0700
- Subject: Re: [PATCH]
- References: <200405040040.CAA06815@faui1d.informatik.uni-erlangen.de>
On Mon, 2004-05-03 at 17:40, Ulrich Weigand wrote:
> Eric Christoper wrote:
> >Having lots of problems with the tpf tracing functions until it was
> >realized that they don't actually affect anything and don't need
> >standard call mechanisms for save and restore either, just execute.
> How come? Your change shouldn't generate any different code;
> could you elaborate what the problems were?
Regrename wanted to start changing things around and inside the branch
and then flow would complain about register liveness later. Optimization
would also play havoc with varargs. I can send you a couple of testcases
if you'd like...
> >+(define_insn "prologue_tpf"
> >+ [(unspec_volatile [(const_int 0)] UNSPECV_TPF_PROLOGUE)]
> >+ "TARGET_TPF"
> >+ "bas\t%%r1,4064"
> >+ [(set_attr "type" "jsr")])
> We should have a (clobber (reg 1)) here at least ...
I thought that it wouldn't be guaranteed to have anything there going in
here, but I can add the clobber.
Eric Christopher <firstname.lastname@example.org>